json_normalize throws AttributeError when calling with array of values

[vuminhle]

Code Sample, a copy-pastable example if possible

xref #21605

from pandas.io.json import json_normalize

json_normalize([1, 2])

Problem description

The above code throws AttributeError:

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "C:\Python36\lib\site-packages\pandas\io\json\normalize.py", line 193, in json_normalize
    for x in compat.itervalues(y)] for y in data]):
  File "C:\Python36\lib\site-packages\pandas\io\json\normalize.py", line 193, in <listcomp>
    for x in compat.itervalues(y)] for y in data]):
  File "C:\Python36\lib\site-packages\pandas\compat\__init__.py", line 211, in itervalues
    return iter(obj.values(**kw))
AttributeError: 'int' object has no attribute 'values'

Expected Output

	0
0	1
1	2

Output of pd.show_versions()

INSTALLED VERSIONS

commit: None
python: 3.6.4.final.0
python-bits: 64
OS: Windows
OS-release: 10
machine: AMD64
processor: Intel64 Family 6 Model 62 Stepping 4, GenuineIntel
byteorder: little
LC_ALL: None
LANG: None
LOCALE: None.None

pandas: 0.23.1
pytest: 3.6.1
pip: 10.0.1
setuptools: 28.8.0
Cython: None
numpy: 1.14.2
scipy: None
pyarrow: None
xarray: None
IPython: 6.3.1
sphinx: None
patsy: None
dateutil: 2.7.2
pytz: 2018.4
blosc: None
bottleneck: None
tables: None
numexpr: None
feather: None
matplotlib: None
openpyxl: None
xlrd: None
xlwt: None
xlsxwriter: None
lxml: None
bs4: 4.6.0
html5lib: 1.0.1
sqlalchemy: None
pymysql: None
psycopg2: None
jinja2: 2.10
s3fs: None
fastparquet: None
pandas_gbq: None
pandas_datareader: None

[WillAyd]

Does this throw an error when dealing with actual JSON objects instead of a list of integers?

[vuminhle]

No. It does not throw if the array contains objects.
However, because an array of values is a valid JSON (https://www.json.org/), I think json_normalize should not throw an error.

[WillAyd]

I'm going to label this as an enhancement then as the documentation for json_normalize states it only works with JSON objects instead of just values.

With that said, I'm -1 on this as I think it just becomes an alternate constructor for a DataFrame. I also don't think this would be supported by read_json (correct if wrong) so it becomes kind of a nuanced corner of our JSON IO ops.

If we did want to do something like this, at the very least I think it should return a Series and not a DataFrame as AFAICT this will never be multi-dimensional (again correct if wrong).

Curious what others think

[gfyoung]

I'm not particularly against having "another constructor" for a DataFrame. That being said, whether we should actually support this I'm not entirely sure.

cc @jreback

[jreback]

so we accept a dict or list of dicts for data. I don't see a reason to expand this scope. However, would take a better error message.

[vuminhle]

Agree with you guys.